Indian Fintech Lending Risk & Growth Model

Business Question — what we set out to answer

Indian digital lending has grown explosively, but credit risk has trailed close behind. This project models the relationship between loan-book growth, gross non-performing assets (GNPA), and cost-to-serve across fintech lenders, identifying which business models deliver growth without blowing up credit quality.

Data Sources — where the data came from

• RBI Financial Stability Reports: Aggregate retail credit and GNPA series.
• Company disclosures & securitisation (PTC) data: Originator-level disbursement and delinquency.
• TransUnion CIBIL industry reports: Segment-wise delinquency benchmarks.

Methodology — how we modelled it

• Book build-up: Top-down and bottom-up build of digital retail loan AUM by product (BNPL, personal, MSME, gold).
• Risk mapping: Vintage-style GNPA curves fitted to portfolio age and origination cohort.
• Unit economics: Cost-to-serve, acquisition cost, and net margin per loan by product line.
• Scenario stress: Rate-rise and unemployment shocks flowing through the risk curves.

Key Insights

• Growth ≠ profitability: The fastest-growing lenders often had the weakest unit economics after reserving for risk.
• Vintage discipline: Lenders who tightened underwriting in FY22–23 now carry materially lower GNPA vintages.
• MSME opportunity vs risk: MSME lending offers the highest spreads but the steepest risk curves.
• Securitisation valve: PTC markets enabled off-balance-sheet growth but concentrated warehousing risk.

Limitations

Public data understates true stress (restructured and sold portfolios move GNPA around). Vintage curves are sensitive to assumptions on collections quality and borrower-level data depth.